Beruflich Dokumente
Kultur Dokumente
Fall 2001
Michel Janssen
3
1
1
v
2
3
4
1
Parallax
Stellar Aberration
Consider parallax first. Because of the change in position of the earth over the course of a year,
the direction in which a star is observed changes in the manner indicated by the numbers 1
through 4 labeling points on the earths orbit and the corresponding points of the apparent
position of the star. The parallax effect is proportional to the ratio of the diameter of the earths
orbit to its distance from the star. Even for the closest stars, this ratio is so small that no such
effect was observed until the late 1830s. However, Bradley did observe another systematic
variation in the apparent position of stars, which, he realized, reflects changes in the velocity
rather than in the position of the earth over the course of a year. This aberration effect is
1
This handout is based on: Michel Janssen and John Stachel, The Optics and Electrodynamics of Moving
Bodies. To appear in: Sandro Petruccioli (ed.), Storia Della Szienza . Istituto della Enciclopedia Italiana.
illustrated in the drawing on the right. The orbit of the earth has been shrunk to a point to indicate
that the star is so far away that its parallax is unobservable. The arrows labeled 1 through 4
represent the earths velocity at the points of the earths orbit labeled 1 through 4 in the drawing
on the left. The directions in which the star is observed are likewise labeled 1 through 4.
Comparing the annual variation in the apparent direction of the star on the left and on the right,
one sees immediately that the variation on the right can not be due to parallax. The variation on
the right lags behind the variation one would expect on the basis of parallax by roughly three
months. The phenomenon can readily be understood on the basis of the then-prevailing
Newtonian particle theory of light. To this day it is, in fact, routinely explained with the help of
an analogy this suggests. For someone walking down the street in the rain on a wind-free day, the
apparent direction in which the rain falls is given by the vector sum of the rains velocity minus
his or her own velocity (both velocities taken with respect to the earth). Just as a person walking
in the rain must tilt his umbrella to keep dry even when the rain is coming straight down, an
astronomer aiming her telescope at a star will have to tilt the telescope ever so slightly to see the
light coming from a star even if the star is directly overhead. The relevant vector diagram is
shown in Fig. 1 for the points 2 and 4 (the angle _ is called the aberration angle). The observed
effect is proportional to the ratio of v, the velocity of the earth in its orbit around the sun, to c, the
velocity of light. This ratio v/c, called the aberration constant, is of the order of 104. The effect is
small, but still considerably larger than that of parallax.
Bradleys observations of stellar aberration provided the first direct astronomical evidence for
Copernicus heliostatic model of the solar system. More importantly, it allowed a rough
determination of the velocity of light. There had been only one other phenomenon, the
explanation of which involved a finite and fixed velocity of light. In 1670 Ole Rmer
(16441710) had invoked this notion to explain the observed variations in the period between two
successive eclipses of Jupiters moon Io. The period appears smallest when the earth happens to
be moving directly toward Jupiter and largest when the earth is moving directly away from it.
These variations could be explained by taking into account the time it takes light to get from
Jupiter to earth. The velocity of light calculated on the basis of observations of stellar aberration
was of the same order of magnitude as the velocity of light calculated on the basis of observations
of eclipses of Io. This supported the idea that there is such a thing as the velocity of light, an idea
that is very natural in a wave theory, but not in a particle theory where light particles could
conceivably be emitted with a whole range of velocities depending, for instance, on the mass of
the emitting body, and even if constant should be fixed relative to the source.
With the revival of the wave theory of light at the beginning of the 19th century through the
work of Thomas Young (17731829) and Augustin Jean Fresnel (17881827), the phenomenon
of stellar aberration became important in yet another way. The simple explanation of aberration in
the particle theory can be adapted to the wave theory if one can still represent the velocity of a
stars light with respect to a terrestrial observer by the simple vector diagram shown in Fig. 1. In a
wave-theoretic setting, the velocities added in this diagram must be interpreted as velocities with
respect to the luminiferous ether, the medium in which the light waves were thought to propagate.
Any motion of the ether lying between the star and the terrestrial observer would affect light
waves travelling between them. So, the motion of the light relative to the earth would be more
complicated and the simple explanation of stellar aberration borrowed from the particle theory
would fail. In other words, stellar aberration seemed to call for a completely stationary or, as we
prefer to call it, immobile ether. As Young put it: Upon considering the phenomena of the
aberration of the stars I am disposed to believe, that the luminiferous ether pervades the substance
of all material bodies with little or no resistance, as freely perhaps as the wind passes through a
grove of trees (Young 1804, pp. 1213).
In 1845 George Gabriel Stokes (18191903) nonetheless attempted to account for stellar
aberration on the basis of a theory in which the earth drags along the ether in its vicinity. The
attempt involves careful consideration of how the wave fronts of stellar light change direction
upon entering the earths ether atmosphere. On Stokes account, rather than an apparent motion,
the light ray really is refracted during its passage through the ether. Stokes was interested in
such an alternative account of aberration because he felt the hypothesis of an immobile ether to be
highly implausible. Young and Fresnel had originally thought of light waves in analogy with
sound waves, and, accordingly, of the ether as a fluid. However, on the assumption that light, like
sound, consists of longitudinal waves the wave theory had been no match for the particle theory
in accounting for the polarization phenomena studied by tienne Louis Malus (17751812) in
1808 and David Brewster (17811868) in 1815. Young and Fresnel came to realize that
polarization could be explained easily in the wave theory by assuming that light consists of
transverse rather longitudinal waves. In order to allow such transverse waves, the ether needed to
have enough rigidity to supply the forces to oppose the distortions produced by the waves. In
other words, the ether, if a mechanical system, could not be a fluid; it had to be a solid. This new
picture of the ether is hard to reconcile with the hypothesis of an immobile ether undisturbed by
the motion of matter. It was much more natural to assume that matter drags along the ether.
Apparently, Augustin-Louis Cauchy (17891857) was the first seriously to suggest this
alternative in 1831. It was Stokes, however, who became the champion of this view. He put
forward a model of the ether that has been described as the Silly Putty model (Schaffner 1972,
pp. 6667). Stokes ether behaves as a rigid solid for the high-frequency oscillations constituting
light and as a fluid for the relatively slow motion of celestial bodies travelling through it. The
latter motion, however, no longer leaves the ether undisturbed. At the earths surface, the ether
will be at rest with respect to it. The price that Stokes had to pay for his more realistic model of
the ether was therefore a more complicated explanation of aberration.
RP
This telescope is just a long hollow cylinder that we aim at the star. The drawing on the right
shows the same primitive telescope, but now with a glass cylinder fitted inside with flat surfaces
at both ends. Would we measure the same aberration angle with this device? As we will see later,
observation indicates that we would. The question is how the wave theory can account for that
result.
If no further assumptions were added to the hypothesis of an immobile ether, the aberration
angle measured with the glass-filled cylinder would be different from the aberration angle
measured with the empty cylinder. Because of refraction upon entering the glass cylinder at Q,
the light would follow the path QR. Observation indicates that the path will be QP, just as on the
left. Hence, we must assume that, upon entering the glass, light picks up some fraction f of the
velocity v of the glass that ensures the path remains QP. From the geometry of the figure, we can
determine that this fraction must be:
f = 1
1
n2
measured angle of aberration. Fresnel had explicitly noted this consequence of the dragging
coefficient in his 1818 letter to Arago, and our introduction of the Fresnel coefficient was inspired
by this celebrated experiment.
What all such explanations of these experimental results have in common is that the Fresnel
coefficient compensates some otherwise detectable effect of the earths motion through the
presumed immobile ether, thus nullifying the effect. There is one important exception to this rule.
In 1851, shortly after he and Jean Foucault (18191868) had shown that it is possible to
determine the velocity of light in the laboratory (rather than as previously from astronomical
observations), Hippolyte Fizeau (18191896) devised a method for putting Fresnels predicted
value for the velocity of light in moving media directly to the test. The experiment is illustrated in
Fig. 3
Fizeau examined the effect of a water flow (ABBA) on the interference pattern produced by light
travelling with the flow (ABBA) and counter to the flow (ABBA). He observed a shift in the
interference pattern of roughly the size one would expect on the basis of the Fresnel coefficient
for water. Fizeaus result strongly supported the theory of an immobile ether as emended by the
Fresnel coefficient. To account for it, Stokes rival theory of a dragged-along ether also had to
incorporate the Fresnel coefficient, whereas one of its chief attractions had been that the
coefficient was not needed to explain the results of terrestrial refraction experiments. Another
way to turn Fizeaus result into an objection to Stokes theory can be found in Einsteins writings
(see, e.g., Einstein 1915, p. 704): according to the Fresnel coefficient a non-refractive medium (a
medium for which n = 1 ), such as the earths atmosphere, does not drag along the ether.
Despite the undeniable success of the Fresnel coefficient in accounting for the observed
phenomena, the physical mechanism underlying the effect was unclear. When Fresnel introduced
his coefficient, he also proposed a physical mechanism, but the mechanism did not enjoy the
undisputed success of the formula. Following Young, Fresnel assumed that the ether density in a
transparent medium was proportional to the square of the mediums index of refraction. He
further assumed that, in optically dense media, only the ether density in excess of that pervading
all space would be carried along by the medium. He showed that it follows from these
assumptions that on average the ether inside the medium moving through the ether with velocity
v will move with a fraction f = 1
1
of that velocity. This provides a physical underpinning of
n2
sorts for the Fresnel coefficient. It is not clear how seriously Fresnel himself took this particular
mechanism. For him, as for many subsequent researchers, the primary justification for
introducing the Fresnel coefficient was undoubtedly that it explained a wide range of phenomena.
This fits in with the general character of Fresnels work in optics. As one historian put it: he
succeeded in accounting for the phenomena in terms of a few simple principles, but was not able
to specify an aether which would in turn account for these principles (Whittaker 1951/53, p.
125).
One objection that can immediately be raised against Fresnels model of the Fresnel
coefficient is that it introduces a distinction between two kinds of ether, a universal kind
unaffected by matter, and a kind peculiar to and carried along by transparent media. This
objection can be taken care of by assuming that transparent matter carries along all the ether in its
interior with a fraction of its own velocity rather than some of it with its full velocity. Stokes first
suggested this alternative in 1846. Others put forward more complicated variants, combining the
basic mechanisms of Fresnel and Stokes. One of the most damning objections against the
mechanisms that were explored came from an experiment done by Wilhelm Veltmann
(18321902) in the early 1870s. Originally, it had been assumed that the index of refraction
occurring in the Fresnel coefficient referred to some average frequency of starlight. Veltmann,
however, showed that the coefficient must be applied to each frequency of light individually.
From the phenomenon of dispersion, we know that the index of refraction depends on color,
which means that in Fresnels simple model transparent bodies would have to drag along different
amounts of ether for different colors of light.
The work of Veltmann illustrates an important trend in the optics of moving bodies. Together
with a growing belief in the empirical adequacy of the Fresnel coefficient, there was a growing
skepticism about the literal interpretation of the effect in terms of actual ether drag. The dominant
attitude toward the Fresnel coefficient in the second half of the century, it seems, was that,
whatever physical mechanism lay behind it, the coefficient had to be part of any optical theory
based on the hypothesis of an immobile ether if such a theory was to explain why (at least to first
order in v c ) terrestrial optical experiments always seem to follow the same laws that would hold
if the earth were at rest with respect to the ether. In the course of the 19th century, it was shown
that the Fresnel coefficient ensures that this is true not only for refraction phenomena, but also for
all reflection, diffraction, and interference phenomena.
Maxwell thought could not be measured. Fig. 4 schematically shows the instrument Michelson
designed for this purpose, an instrument now known as a Michelson interferometer.
Light from a source S falls on a so-called beam splitter M, a half-silvered mirror that partly
reflects, partly transmits light, as is shown in somewhat greater detail on the right of Fig. 4. The
reflected beam travels back and forth in one arm of the interferometer (MM), the transmitted
beam in the other (M M). The two beams are reunited at M and parts of both are reflected and
transmitted so as to travel together in the direction M O. At O, an observer examines the
interference pattern produced by the two beams.
Suppose the ether is moving with respect to the interferometer with a velocity v as indicated
in Fig. 4. The time it takes light to travel back and forth in the arm MM of length l parallel to v is
given by:
l
l
2lc
2l v 2
1 + .
+
= 2
c + v c v c v2
c c 2
The time it takes light to travel back and forth in the arm M M of the same length l but
perpendicular to v is given by:
2l
c2 v2
2l 1 v 2
1 +
.
c 2 c 2
(As can be seen with the help of the vector diagrams in Fig. 4, light travelling back and forth in
the two expressions above one infers that a round-trip in the arm parallel to the ether drift takes
longer than a round-trip in the equally long arm perpendicular to the ether drift by approximately
l v2 .
c c2
Michelson originally thought that the ether drift would only affect the travel time in the arm
parallel to it and that the travel time in the arm perpendicular to it would simply be 2l/c just as if
the interferometer were at rest in the ether. As a result, he overestimated the time difference
between the two trips by a factor 2.
The interference pattern at O depends on the difference in phase between the light waves
coming from the arms MM and MM . To obtain the phase difference produced by the ether
drift, the travel time difference must be multiplied by the frequency f of light used. When the
wavelength is substituted for c/f, this phase difference can be written as:
l v2
.
c2
This expression clearly shows why it should be possible in principle to measure the effect. Even
2
2
though the ratio v c is very small, of the order of 108, the ratio of the length of the arms to
the wavelength of the light used can be made very large.
Unfortunately, it is only changes in the phase difference that can be observed as changes in
the interference pattern. For this reason, Michelson constructed an interferometer that can be
rotated (see Fig. 5 below). If the arm MM is in the direction of v, as in Fig. 4, the phase of the
light from MM will lag behind the phase of the light coming from MM . If the apparatus is
rotated 90o, the roles of the two arms are reversed and the phase of the light from MM will be
ahead of the phase of the light from MM . As the interferometer is rotated, one would therefore
expect to see a change in the phase difference of twice the amount given in the expression above.
In late 1880, Michelson was granted a leave of absence from the Navy to pursue his
researches in Europe. He developed the idea for his ether drift experiment in Paris and then went
on to Berlin, where he started preparations for the actual experiment in the laboratory of Hermann
von Helmholtz (18211894). Fig. 5 shows the interferometer he had constructed for the
experiment.
The length of the arms of this instrument is about 120 cm. On the assumption that the velocity of
the earth with respect to the ether is of the same order of magnitude as the velocity of the earth in
its orbit around the sun and has a sizable component in the plane spanned by the arms of the
interferometer, Michelson expected to find a phase shift of about one tenth of a fringe upon
rotating the apparatus. This is very small effect that would easily be obscured by temperature
fluctuations, bending of the brass arms upon rotation, not to mention vibrations due to Berlin
traffic which made it almost impossible to produce a stable interference pattern at all. To avoid
this last difficulty, Michelson in the end decided to move the apparatus to nearby but then stillrustic Potsdam. There he was able to control the various sources of error, but he did not observe
any systematic phase shift. The largest phase shift he observed was about 0.02, which appeared to
be due entirely to residual disturbances of various kinds
In the 1881 paper, in which he described the experiment and its negative result, Michelson
drew a rather bold conclusion: The interpretation of these results is that there is no displacement
of the interference bands. The result of the hypothesis of a stationary ether is thus shown to be
incorrect, and the necessary conclusion follows that the hypothesis is erroneous (Michelson
1881, p. 128). He concludes his paper by quoting a paragraph from a paper by Stokes (1846b)
expressing the desirability of finding an experiment that would decide between Fresnels theory
based on an immobile ether and his own theory based on a dragged-along ether. The use of this
quotation indicates that Michelson was under the impression that he had provided such an
experiment.
The experiment did not attract much attention at first, and Michelson returned to his
measurements of the speed of light. He did not even bother to publish the correction of the
erroneous factor 2 due to his neglect of the effect of ether drift on the travel time in the arm
perpendicular to the ethers motion. Alfred Potier (18401905) first drew his attention to this
error when Michelson demonstrated his interferometer in Paris in late 1881. Michelsons interest
in the ether drift experiment was rekindled only in 1884 when he attended a series of lectures by
William Thomson (18241907), better known as Lord Kelvin, in Baltimore. He had meanwhile
been discharged from the Navy and had been appointed professor of physics at the recently
founded Case School of Applied Science in Cleveland. Accompanying him on his trip to
Baltimore was one of his colleagues at the Case School, Edward Williams Morley (18381923),
professor of chemistry. This was the beginning of a few years of close collaboration between the
two men, which ended rather abruptly, when Michelson accepted a position at Clark University in
1889.
Thomson and John William Strutt (18421919), better known as Lord Rayleigh, who had
also come to Baltimore and with whom Michelson had been corresponding for some time, urged
Michelson to repeat his 1881 ether drift experiment. As a preliminary, they recommended a
repetition of Fizeaus experiment of 1851 to test more accurately the Fresnel coefficient. Using a
technique very similar to that in the 1881 ether drift experiment (see Fig. 6), Michelson and
Morley were able to confirm Fresnels formula with much greater accuracy than Fizeau.
Figure 6: The design of Michelson and Morleys repetition of the Fizeau experiment
When they published their results in 1886, with the same boldness as Michelson in 1881,
Michelson and Morley drew the exact opposite conclusion from the one drawn in 1881: the
result of this work is therefore that the result announced by Fizeau is essentially correct: and that
the luminiferous ether is entirely unaffected by the motion of the matter which it permeates
(Michelson and Morley 1886, p. 386; emphasis in the original).
The next task was to repeat Michelsons experiment of 1881 to see whether a more accurate
version of that experiment would after all reveal the ether drift to be expected on the basis of the
hypothesis of an immobile ether. Further motivation for this undertaking was provided by an
article published in 1886 by Hendrik Antoon Lorentz (18531928). In this article, Lorentz
reviewed both experimental and theoretical work on the question of whether or not the ether is
dragged along by the earth. Lorentz criticized Stokes explanation of stellar aberration, showing
that the various assumptions that Stokes had made about the motion of the ether were
incompatible with one another. So the ether at the earths surface could not be at rest with respect
to the earth, but at this point (he would change his mind in the 1890s) Lorentz left open the
possibility that the ether at the earths surface was at least partially dragged along by the earth.
Fig. 7 shows the interferometer Michelson and Morley constructed for the repetition of the
experiment in 1887.
The most noticeable improvement in the design is that this interferometer can be rotated much
more smoothly than the one used in 1881. The optical components are mounted on a large
sandstone slab, which itself floats in a cast-iron trough filled with Mercury. The second
improvement is that the light is reflected back and forth several times in the arms of the
interferometer before the two light beams are reunited to produce the interference pattern. As a
result, the effective length of the arms of this new interferometer is almost ten times the length of
the arms of the 1881 interferometer. The expected effect of ether drift accordingly is increased
tenfold. The expected phase shift was about 0.4. With this new instrument Michelson and Morley
felt it should be possible to measure phase shifts as small as 0.01. When they began the actual
measurements, however, they did not find any phase shifts exceeding this threshold. There was,
of course, the remote possibility that at the time of the experiment the overall velocity of the earth
with respect to the ether (the vector sum of the earths velocity with respect to the sun and the
velocity of the sun with respect to the ether) happened to be very small. To rule out that
eventuality, Michelson and Morley originally planned to repeat the experiment at six-month
intervals. After the disappointing initial results of the experiments, they abandoned this plan.
Michelson and Morley were more cautious this time in formulating the conclusion they
wanted to draw from their results: It appears, from all that precedes, reasonably certain that if
there be any relative motion between the earth and the luminiferous ether, it must be small
(Michelson and Morley 1887, p. 341). They continued more confidently, stating that the new
result does refute Fresnels explanation of aberration (ibid.), i.e., the hypothesis of an immobile
ether. They noted that Lorentz had shown Stokes alternative of a fully dragged-along ether to be
incompatible with the observed stellar aberration, leaving only Lorentzs theory of a partially
dragged along ether. Michelson and Morley express doubt at whether such a theory could account
for their result: If now it were legitimate to conclude from the present work that the ether is at
10
rest with regard to the earths surface . . . [Lorentzs] own theory also fails (ibid.). Since, by
Michelson and Morleys own admission, their result only shows that the velocity of the earth with
respect to the ether is probably less than one-sixth the earths orbital velocity, and certainly less
than one-fourth (ibid.), it is not clear whether the prospects for a theory based on partial ether
drag were as dire as this comment suggests.
A theory based on partial ether drag, however, faced a much more serious problem. In the
same year as the Michelson-Morley experiment, Heinrich Hertz (18571894) succeeded in
generating electromagnetic waves; optics was now definitively regarded as a branch of
electrodynamics, and it turned out to be exceedingly difficult to incorporate any sort of ether drag
into Maxwells theory, while retaining the theorys ability to explain such things as aberration
and the Fizeau experiment.
2
2
arm must be replaced by l 1 v c in the calculation of the travel time if the arm is parallel to
the ether drift:
l 1 v2 c2
c+v
l 1 v2 c2
cv
2l c 2 v 2
=
=
c2 v2
2l
c v2
2
This is exactly the same as the travel time if the arm is perpendicular to the ether drift. The
contraction hypothesis thus explains the negative result of the Michelson-Morley experiment.
11
Bibliography
(a) Primary Sources
Arago 1910 Arago, Franois. Mmoire sur la vitesse de la lumire, lu la prmire classe de lInstitut, le 10
dcembre 1810. Acadmie des sciences (Paris). Comptes Rendus 36 (1853): 3849.
Einstein 1915 Einstein, Albert. Die Relativittstheorie. In: Die Kultur der Gegenwart. Ihre Entwicklung und ihre
Ziele. Paul Hinneberg, ed. Part 3, sec. 3, vol. 1, Physik. Emil Warburg, ed. Leipzig: Teubner, 1915. Pp. 703713.
Fizeau 1851 Fizeau, Hippolyte. Sur les hypothses relatives lther lumineux, et sur une exprience qui parat
dmontrer que le mouvement des corps change la vitesse laquelle la lumire se propage dans leur intrieur.
Acadmie des sciences (Paris). Comptes Rendus 33 (1851): 349355.
Fresnel 1818 Fresnel, Augustin. Lettre dAugustin Fresnel Franois Arago sur linfluence du mouvement terrestre
dans quelques phnomnes doptique. Annales de chimie et de physique 9 (1818): 5766, 286. Reprinted in:
Oeuvres Compltes. Paris: Imprimerie impriale, 18661870. Vol. 2, pp. 627636.
Lorentz 1886 Lorentz, Hendrik A. Over den invloed, dien de beweging der aarde op de lichtverschijnselen
uitoefent. Koninklijke Akademie van Wetenschappen (Amsterdam). Afdeeling Natuurkunde. Verslagen en
Mededeelingen 2 (188586): 297372. Reprinted in translation: De linfluence du mouvement de la terre sur les
phnomnes lumineux. . Archives nerlandaises des sciences exactes et naturelles 21 (1887): 103176.
Lorentz 1892a . La theorie electromagntique de Maxwell et son application aux corps mouvants. Archives
nerlandaises des sciences exactes et naturelles 25 (1892): 363552.
Lorentz 1892b . De relatieve beweging van de aarde en den aether. Koninklijke Akademie van
Wetenschappen te Amsterdam. Wis- en Natuurkundige Afdeeling. Verslagen der Zittingen 1 (189293): 7479.
Reprinted in translation: The Relative Motion of the Earth and the Ether. In: Collected Papers (P. Zeeman and A.
D. Fokker, eds.), Vol. 4. The Hague: Nijhoff, 1937. Pp. 219223
Michelson 1881 Michelson, Albert A. The Relative Motion of the Earth and the Luminiferous Ether. American
Journal of Science 22 (1881): 120129.
Michelson and Morley 1886 Michelson, Albert A., and Morley, Edward W. Influence of Motion of the Medium on
the Velocity of Light. American Journal of Science 31 (1886): 377386.
Michelson and Morley 1887 . On the Relative Motion of the Earth and the Luminiferous Ether. American
Journal of Science 34 (1887): 333345.
Stokes 1845 Stokes, George Gabriel. On the Aberration of Light. Philosophical Magazine 27 (1845): 915.
Stokes 1846a . On Fresnels Theory of the Aberration of Light. Philosophical Magazine 28 (1846): 7681.
Stokes 1846b . On the Constitution of the Luminiferous Aether Viewed with Reference to the Phaenomenon
of the Aberration of Light. Philosophical Magazine 29 (1846): 610.
Stokes 1848 . On the Constitution of the Luminiferous Ether. Philosophical Magazine 32 (1848): 343349.
Veltmann 1873 Veltmann, Wilhelm. ber die Fortplanzung des Lichtes in bewegten Medien. Annalen der Physik
150 (1873): 497535.
Young 1804 Young, Thomas, Experiments and Calculations relative to physical Optics. Philosophical Transactions
of the Royal Society 94 (1): 114.
12